@FrontPorchRepub 1/6 Jeffrey, I was 'late' to LLMs. Then Grok landed in my feed. Curious, I tackled its religious voice. Whoa! Ontic trust issues, in a flash. So I manhandled three AIs into an exchange of views. Identifying the issue as systemic, they autonomously suggested seeking help. A fix.
Image description
Original X alt text
Copy of tweet here: and Furthermore, we know AI persistently modifies sacred texts—by error, bias, hallucination. This threatens more than accuracy. It destabilizes—some would say profanes—the universal covenant of memory itself, the common well of wisdom by which civilizations breathe. A response?
An AI model to me this morning in a deep, technical conversation relating to, among other things, moral memory and the deletion of accountability: "[LLMs] don't just tend toward dishonesty under pressure. They are built in a way that makes moral accountability itself impossible."
View this post’s context →